Skip to content

Home

Jump to bottom
ShweataNHegde edited this page Jun 12, 2021 · 35 revisions

Welcome to the CEVOpen wiki! This page outlines the key components of the project. It's intentionally kept short. If you wish to know more, you can browse through the wiki pages of CEVOpen and openVirus.

1. Main components of intern activity:

Currently, our projects are based on building dictionaries. Each intern has a dictionary or dictionaries, which is usually relevant to essential oils.

1.1 Interns and Roles

Interns Dictionary Mini-Project Other Roles Date of Joining Date of Leaving
Radhu Kantilal Ladani @Radhu903 eoPlant, activity Medicinal Activity Record keeping 04-01-2021 30-06-2021
Kanishka Parashar @Kanishkaparashar03 invasive_plant Invasive Species 11-01-2021 30-06-2021
Talha Hasan @tlhahsn plant_compound, eoplant_material_history Plant Chemical Compounds 01-02-2021 31-07-2021
Vasant Kumar @950vasant eoplant_part, gene Plant Genes 01-02-2021 31-07-2021
Dr. Sagar Sudam Jadhav @sasujadhav1 Dictionary Manager 18-05-2021 31-10-2021
Chaitanya Kumar @chaitshar Record keeping (coding sessions) 01-06-2021 01-08-2021
Bhavani Malhotra @malhotra-bhavini 25-05-2021 30-09-2021
Ayush Garg (Volunteer) @ayush4921 Developing pygetpapers
Shweata N. Hegde @ShweataNHegde organization, plant_genus, ethics_statement Ethics Statement Project Manager, Record Keeping, Outreach 01-06-2021 01-09-2021

1.2. Mini-projects

  • chemotype
  • genotype
  • activities (medicinal)
  • phenotype - invasive species integration - how these fit together - an atlas

2. Technology

Tools include:

  • APIs for repositories such as EPMC, biorXiv preprints, and thesis collections.
  • Scrapers for semi-structured sites such as journals
  • standardised metadata (e.g. JATS)
  • PDF and HTML readers => XML or JSON
  • article sectioning (e.g. into JATS categories)
  • extraction of floats (tables, maps, images, diagrams, chemistry, maths*)
  • display and navigation of sections in a paper
  • aggregated statistics and machine learning
  • multilingual annotation (using Wikidata)
  • linking to the Wikidata knowledge graph

2.1.((py)getpapers, ami)

  • pygetpapers is the scraper developed in Python by Ayush Garg. It is based on getpapers(https://github.com/ContentMine/getpapers) which was written in Node.js. pygetpapers downloads scientific papers, primarily from EuropePMC repository. You can read more about it here
  • pyami (Needs more documentation. Still a prototype) is currently being developed by Peter Murray-Rust. It's a new Python-based open-source universal reader and analyser for scientific literature. Source code can be found here

2.2. How are the dictionaries created?

  • Most dictionaries are created from Wikidata SPARQL queries. You can take a look at individual dictionary wiki pages to know more.
  • You can also refer to this slide deck to understand the basics.

2.3. Reporting Errors

When you have problems please try to be as constructive and informative as possible. Mailing developers (PMR or Ayush) with "It doesn't work, please tell me what to do" is not only useless.

RULE. No one can fix your bug unless you describe it. So, describe it:

  • What are you trying to do? In precise terms: "I am searching corpus X using program Y according to instructions Z"
  • What is your environment?" "I am on a Windows 10 machine with 16 GB RAM"
  • Is this a new problem? "No, it has just appeared today". It worked yesterday
  • Can you reproduce it? Yes. DO OTHER PEOPLE REPRODUCE THE SAME ERROR?
  • what did you do? I ran myprog on the command line
myprog -boo mydata.dat
  • what happened? I immediately got this stack trace:
<output>
<stack trace goes here>
  • ONLY use screenshots when it's a graphical program. Never use photographs.
  • Were there any other messages?
"cannot find data file:" myfile.dat

3. Prerequisites

Python is essential to run all of our software. Ensure you've installed it before proceeding further.

3.1.Install

3.1.1. pygetpapers (https://github.com/petermr/pygetpapers)

Run the following command on your command line to install pygetpapers

pip install git+git://github.com/petermr/pygetpapers

If you have trouble installing using this method, you can find alternatives here.

3.1.2. ami_gui.py

  • git clone https://github.com/petermr/openDiagram.git
  • Though ami_gui.py runs on the command line, you will have to make some changes to the source code to point the software to where all the projects outlined below lie on your local machine. PyCharm is recommended to edit the source code.

3.2. git clone

The project has gradually expanded and branched out to different research areas. Therefore, our work is dispersed across various different repositories. These repositories are where the latest dictionaries, mini-corpora and software are. To run amigui_py, you will have to clone (i.e., download it to your local machine) the following repositories:

4. Overall Goal

To build a multilingual semantic Atlas of Volatile Phytochemistry.[1]

4.1. Subgoals

To build Open Source multiplatform tools which can discover, aggregate, clean, and semantify scholarly documents containing significant amounts of phytochemical VOC[2]s. Documents will contain, extraction and assay of oils, optionally with properties and activities.

4.2. About CEVOpen

Phytochemistry is the key component of this project and in the main, we will be analysing:

  • compounds (mainly VOC). Includes synonyms, structures, images
  • plants that create VOC/essential oils, again many synonyms, includes images
  • locations where the plant was harvested
  • activities reported for the oils
  • organizations involved

We will be analysing corpora for instances of the above, manually to validate the process and then automatically.

[*] not included in CEVOpen but extensible in future
[1] we need an engaging title. "Atlas" is often extended beyond maps (e.g. Atlas of The Human Body). For example, plantPart is an atlas of the plant. It works for me but may confuse others. Here are some ideas:

  • "Compendium of ..."
  • "Semantic Essence of phytochemistry". Essence == central meaning, and also volatiles
    But please think creatively.

[2] Volatile Organic Compound

4.4. Required actions:

  • Coordination of EO-related and general dictionaries - conformance to a common standard.
  • Validation of gold-standard minicorpora (e.g. for training and validating machine learning)
  • If you are interested in contributing to the project on the Machine Learning front, you can take a look at the Our-Project-and-Machine-Learning page.

4.5. Update (2021-06-06):

We have a new set of interns joining us. Here we are summarizing goals for the next 6 months:

4.5.1. Goals and Objectives

We have been joined by Chaitanya Sharma and Bhavini Malhotra and Sagar Jadhav and we are hoping to appoint another intern (InternX) shortly. The goal of these 6 months is to consolidate our current dictionaries, corpora, and code, and then to explore how they can be used. We'll think of this as a guide to phytochemistry of essential oils ("Atlas", "Compendium", etc.). Each of you will be creating a specific part of this and/or coordinating and customising it for a wide range of audiences. High-level objectives: - which will need prioritising

  1. create
    • tools and
    • knowledge for phytochemistry of essential oils
  2. carry out initial scoping research
  3. create outreach materials and events

4.5.2. Roles

We have just about enough tools to start semi-automating the process, but they will need refining. That's the roles of:|

  • Chaitanya - section and entity recognition in papers (including PDFs). This is critical for scaling up and extracting paragraphs and sentences related to all our dictionaries
  • Bhavini - presentation and analysis of data. This is downstream, so consumes AMI output and hopefully will reveal potential patterns.This can lead to a "Phytochemical Atlas" linking plants, chemistry and geography. Both B and C have good software experience and will help with code testing, documentation and maintenance
  • Shweata will be needed in coordinating this, writing linking material, etc. Her work on Ethics Statement will support text analysis and may link up with Chaitanya. She can also work with Bhavini on presentation. All of these are transferable skills. Sagar will build a plant_gene/gene_ dictionary. This is more speculative than previous dictionaries since the literature is more dispersed and variable than the EO->compounds->activity papers we have found so far. Also, the nomenclature is more challenging. He will also soak-test (sic, not smoke) the current dictionaries and (manually at first) validate the results.
  • We expect the new intern (X) to have a range of skills in both informatics and bioscience. A multidisciplinary approach rather than concentrating on one area of science. We need someone to help pull the data together, clean it and look for patterns. A disciplined worker, good communication skills, flexible, competent in data handling etc. rather than specific bio/chemical knowledge. All will be heavily involved in testing the AMI framework including pygetpapers. Those experienced in the software will help to make the choice of systems we use.

5. Meeting Records

We regularly meet bi-weekly to discuss strategy and work. Apart from that, we also meet informally to review code - once a week.

5.1. Biweekly Meetings

Radhu had maintained records of all meetings since the beginning of her term. https://github.com/petermr/dictionary/wiki/Meeting-Records

Bhavini is maintaining the recordings starting 2021-06-07.

5.2. Coding Sessions

We are now having regular coding sessions every week. Chaitanya is going to maintain records of these meetings. https://github.com/petermr/CEVOpen/wiki/Coding-Sessions:-Meeting-Record

6. Outreach

We've presented our work (mostly of openVirus) at various places including Wikcite, COAR and BarCamp. You can take a look at our Outreach page. If you're a newbie, taking a look at our presentations is, probably, the best way to get started to understand the pipeline.

7. Code of Conduct

All the interns, volunteers and contributors should adhere to the code of conduct, outlined here. Basically, it says "be respectable and helpful towards everyone".

Clone this wiki locally